A method and device for testing and diagnosing energy storage devices based on multi-model fusion and AI collaboration

Abstract

This invention discloses a method and apparatus for detecting and diagnosing energy storage devices based on multi-model fusion and AI collaboration. The method starts a portable host, drives the detection modules to perform self-tests, initializes the AI ​​processing module, and establishes a connection by scanning surrounding devices using the communication module. Voice commands control each detection module to collect multi-dimensional raw data. Each diagnostic model quantitatively calculates and outputs single-dimensional diagnostic results and confidence levels. A weighted fusion of these results using a self-built core fusion model yields a comprehensive diagnostic value, determining the fault level and type to generate a comprehensive diagnostic result. Protocols are parsed as needed to generate adaptation code, supplementing the collection of auxiliary data. The optimal operation and maintenance solution is selected based on the comprehensive diagnostic result using formula calculations, outputs the solution, and drives its execution. Data is simultaneously uploaded to the cloud. Finally, model parameters and weight coefficients are updated based on the device's operating data after fault handling, improving subsequent diagnostic accuracy. This invention solves the problems of difficult cross-manufacturer interface adaptation and inefficient operation in extreme environments found in existing technologies.

Description

Technical Field

[0001] This invention relates to the field of energy storage equipment testing and fault diagnosis technology, specifically to a method and device for energy storage equipment testing and diagnosis based on multi-model fusion and AI collaboration. Background Technology

[0002] Currently, energy storage systems are widely used in scenarios such as new energy grid connection, grid peak shaving, and industrial and commercial backup power. With the large-scale deployment of energy storage equipment, the coexistence of equipment from multiple manufacturers and applications in extreme environments are becoming increasingly common. Existing testing technologies mostly rely on single-dimensional equipment, and some integrated testing devices can only achieve basic data collection. Moreover, they require manual writing of adaptation code to be compatible with the interface protocols of different manufacturers. In actual operation and maintenance, they are gradually becoming the mainstream testing auxiliary tools in the industry.

[0003] However, existing technologies have significant bottlenecks: on the one hand, multi-dimensional detection data are fragmented and lack quantitative fusion analysis based on mathematical models, making it easy to miss cross-system related faults; on the other hand, the interface protocols of equipment from different manufacturers vary greatly, resulting in long and inefficient manual adaptation cycles, and manual operation and code writing in the cold northern environment are easily affected by low temperatures, leading to delayed fault response. Furthermore, the diagnosis of problems such as liquid cooling system leakage and battery degradation relies on experience-based judgment, lacking accurate quantitative calculation models, making it difficult to output globally optimal operation and maintenance solutions, and failing to meet the needs of efficient and intelligent detection and operation and maintenance of energy storage equipment.

[0004] Therefore, there is an urgent need for a detection and diagnosis method for energy storage devices based on multi-model fusion and AI collaboration to solve the problems of fragmented multi-dimensional detection, difficulty in cross-manufacturer interface adaptation, and inefficiency in extreme environment operation of existing technologies. Summary of the Invention

[0005] To address these issues, this invention provides a method and apparatus for detecting and diagnosing energy storage devices based on multi-model fusion and AI collaboration. This method solves problems such as fragmented multi-dimensional detection, difficulty in cross-manufacturer interface adaptation, inefficiency in extreme environment operation, and lack of accurate quantitative diagnosis and globally optimal operation and maintenance solutions in existing energy storage device testing.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for detecting and diagnosing energy storage devices based on multi-model fusion and AI collaboration, characterized in that it includes: The portable host is started, driving each detection module to complete self-test; the AI ​​processing module loads the preset diagnostic model and pre-trained data of the manufacturer's interface protocol; the communication module establishes the basis for device connection by scanning the surrounding energy storage devices; Based on the aforementioned device connectivity, the camera image recognition module, infrared temperature detection module, pressure detection module, and active equalization battery detection module are controlled via voice commands to simultaneously collect images of the water pipe connection, device surface temperature, liquid cooling system pressure and level, and individual battery parameters, thereby obtaining multi-dimensional raw data. Based on the multi-dimensional raw data, each diagnostic model is quantitatively calculated using corresponding mathematical formulas, and outputs single-dimensional diagnostic results and model confidence respectively. Based on the single-dimensional diagnostic results and the model confidence level, a weighted fusion is performed through a self-built core fusion model to obtain a comprehensive diagnostic value; based on the comprehensive diagnostic value, the fault level and type are determined, and a comprehensive diagnostic result is generated. Based on the supplementary data requirements of the comprehensive diagnostic results, the AI ​​processing module parses the target energy storage device interface protocol through an artificial intelligence code editor and automatically generates adaptation code; based on the adaptation code, it establishes communication with the target energy storage device and EMS system to collect supplementary auxiliary data. Based on the comprehensive diagnostic results and the auxiliary data, the globally optimal operation and maintenance solution is calculated and screened using the comprehensive loss value formula. The globally optimal operation and maintenance plan is output to the display and alarm module; based on the globally optimal operation and maintenance plan, the testing personnel execute auxiliary operations through voice command-driven devices, and upload fault data, diagnostic process and execution status to the cloud platform; Based on the equipment operation data after fault handling, the preset parameters and confidence weight coefficients of the diagnostic model are updated to improve the accuracy of subsequent diagnosis.

[0007] As a preferred solution for the detection and diagnosis method of energy storage devices based on multi-model fusion and AI collaboration, the diagnostic models include: infrared temperature recognition model, image recognition model, stress test model, active balancing battery management system model and EMS energy management system model.

[0008] As a preferred solution for the detection and diagnosis method of energy storage equipment based on multi-model fusion and AI collaboration, after the camera image recognition module collects data, the image recognition model calculates the single-dimensional diagnostic result through the feature point matching similarity formula. The formula for feature point matching similarity is:

[0009] In the formula, S Match feature points with similarity; N This represents the number of feature points that successfully match the image to be detected with the standard image. This represents the total number of standard feature points corresponding to the connector.

[0010] As a preferred solution for the detection and diagnosis method of energy storage equipment based on multi-model fusion and AI collaboration, after the infrared temperature detection module collects data, the infrared temperature recognition model calculates the single-dimensional diagnosis result by sequentially using the temperature correction formula, the eddy current heating judgment formula, and the poor contact heating judgment formula. The temperature correction formula is as follows:

[0011] In the formula, T This is the corrected temperature; The raw temperature collected by the infrared sensor; The ambient temperature; k This is an environmental correction factor; The formula for determining eddy current heating is as follows:

[0012] In the formula, Temperature rise in equipment components caused by eddy current effect; I This is the rated current of the line; d The outer diameter of the line; The formula for determining heat generation due to poor contact is as follows:

[0013] In the formula, The contact resistance of the line or plug; This represents the temperature rise difference between the plug and the wiring. This represents the thermal resistance of the corresponding component.

[0014] As a preferred solution for the detection and diagnosis method of energy storage equipment based on multi-model fusion and AI collaboration, after the pressure detection module collects data, the pressure test model performs liquid leakage diagnosis of the liquid cooling system through the pressure change rate formula and leakage estimation formula; and performs liquid shortage diagnosis of the liquid cooling system through the low liquid judgment formula and the optimal liquid replenishment formula. The formula for the rate of change of pressure is:

[0015] In the formula, The pressure change rate of the liquid cooling system; , The measured pressures of the liquid cooling system at different times; , For the corresponding , The detection time; The formula for estimating the leakage amount is:

[0016] In the formula, This refers to the amount of leakage. V Volume of the liquid cooling system; The formula for determining low fluid volume is:

[0017] In the formula, h Real-time liquid level of the liquid cooling system; This represents the initial liquid level of the liquid cooling system. This represents the theoretical change in liquid level. A This represents the cross-sectional area of ​​the liquid storage chamber in the liquid cooling system. The formula for the optimal fluid replacement volume is:

[0018] In the formula, This represents the optimal replenishment volume for the liquid cooling system. This is the rated optimal liquid level for the liquid cooling system.

[0019] As a preferred solution for the detection and diagnosis method of energy storage devices based on multi-model fusion and AI collaboration, after the active balancing battery detection module collects data, the active balancing battery management system model calculates the single-dimensional diagnostic result through the battery consistency evaluation formula, the battery degradation diagnosis formula and the optimal balancing current formula. The formula for evaluating battery consistency is as follows:

[0020] In the formula, For battery pack SOC balance; This represents the maximum SOC (State of Charge) of a single cell in the battery pack. This represents the minimum SOC (State of Charge) of a single cell in the battery pack. This represents the average SOC of each cell in the battery pack. The battery degradation diagnosis formula is as follows:

[0021] In the formula, For the first i The internal resistance decay rate of each battery cell; For the first i The current measured internal resistance of each individual battery cell; For the first i The initial internal resistance of each battery cell; The formula for the optimal equilibrium current is:

[0022] In the formula, For the first i The active balancing current of each battery cell; This is the equilibrium coefficient.

[0023] As a preferred solution for the detection and diagnosis method of energy storage equipment based on multi-model fusion and AI collaboration, in the process of obtaining the comprehensive diagnostic value by weighted fusion through the self-built core fusion diagnostic model, the weight of each diagnostic model is determined by the confidence formula; based on the weight of each diagnostic model, the comprehensive diagnostic value is calculated by the weighted fusion formula. The confidence level formula is as follows:

[0024] In the formula, For the first i The weights of each diagnostic model; For the first i Diagnostic accuracy of a diagnostic model; i =1 corresponds to the infrared temperature recognition model; i =2 corresponds to the image recognition model; i =3 corresponds to the active equalization battery management system model; i =4 corresponds to the EMS energy management system model; The weighted fusion formula is as follows:

[0025] In the formula, D This is a comprehensive diagnostic value; For the first i A single-dimensional diagnostic value output by a diagnostic model.

[0026] As a preferred solution for the detection and diagnosis method of energy storage equipment based on multi-model fusion and AI collaboration, when the AI ​​processing module parses the interface protocol, the adaptation effect is determined by the interface protocol matching degree formula. The formula for the interface protocol matching degree is:

[0027] In the formula, M P represents the protocol matching degree; P is the interface protocol feature vector of the energy storage device to be adapted. The first one pre-stored in the database j Interface protocol feature vectors of each manufacturer; In the process of calculating and selecting the globally optimal operation and maintenance solution using the comprehensive loss value formula, the comprehensive loss value formula is as follows:

[0028] In the formula, L This is the overall loss value; The estimated processing time for the operation and maintenance plan; This is the estimated processing cost for the operation and maintenance plan.

[0029] This invention also provides an energy storage device detection and diagnosis apparatus based on multi-model fusion and AI collaboration, employing the above-mentioned energy storage device detection and diagnosis method based on multi-model fusion and AI collaboration, including: The power-on initialization unit is used to start the portable host and drive each detection module to complete self-test; the AI ​​processing module loads the preset diagnostic model and pre-trained data of the manufacturer's interface protocol; the communication module establishes the device connection foundation by scanning the surrounding energy storage devices. The multi-dimensional data acquisition unit is used to control the camera image recognition module, infrared temperature detection module, pressure detection module and active equalization battery detection module through voice commands based on the device connection foundation, and simultaneously acquire the line water pipe connection image, equipment surface temperature, liquid cooling system pressure and liquid level and battery pack individual parameter data to obtain multi-dimensional raw data; The single-dimensional model diagnostic unit is used to perform quantitative calculations on each of the diagnostic models based on the multi-dimensional raw data using corresponding mathematical formulas, and output single-dimensional diagnostic results and model confidence scores respectively. The multi-model fusion diagnostic unit is used to perform weighted fusion based on the single-dimensional diagnostic results and the model confidence level through a self-built core fusion model to obtain a comprehensive diagnostic value; based on the comprehensive diagnostic value, the fault level and type are determined, and a comprehensive diagnostic result is generated. A cross-manufacturer interface adaptation unit is used to supplement data requirements based on the comprehensive diagnostic results. The AI ​​processing module parses the target energy storage device interface protocol through an artificial intelligence code editor and automatically generates adaptation code. Based on the adaptation code, communication is established with the target energy storage device and EMS system to supplement and collect auxiliary data. The optimal operation and maintenance plan generation unit is used to calculate and screen the globally optimal operation and maintenance plan based on the comprehensive diagnostic results and the auxiliary data, using the comprehensive loss value formula. The solution execution and data upload unit is used to output the global optimal operation and maintenance solution to the display and alarm module; based on the global optimal operation and maintenance solution, the testing personnel execute auxiliary operations through voice command-driven devices and upload fault data, diagnostic process and execution status to the cloud platform. The model optimization and update unit is used to update the preset parameters and confidence weight coefficients of the diagnostic model based on the equipment operation data after fault handling, so as to improve the accuracy of subsequent diagnosis.

[0030] As a preferred solution for energy storage device detection and diagnosis based on multi-model fusion and AI collaboration, the diagnostic models in the power-on initialization unit include: infrared temperature recognition model, image recognition model, stress test model, active balancing battery management system model, and EMS energy management system model.

[0031] As a preferred solution for a detection and diagnosis device for energy storage equipment based on multi-model fusion and AI collaboration, in the single-dimensional model diagnosis unit, after the camera image recognition module collects data, the image recognition model calculates the single-dimensional diagnosis result through a feature point matching similarity formula. The formula for feature point matching similarity is:

[0032] In the formula, S Match feature points with similarity; N This represents the number of feature points that successfully match the image to be detected with the standard image. This represents the total number of standard feature points corresponding to the connector.

[0033] As a preferred solution for energy storage equipment detection and diagnosis device based on multi-model fusion and AI collaboration, in the single-dimensional model diagnosis unit, after the infrared temperature detection module collects data, the infrared temperature recognition model calculates the single-dimensional diagnosis result by sequentially using the temperature correction formula, the eddy current heating determination formula, and the poor contact heating determination formula. The temperature correction formula is as follows:

[0034] In the formula, T This is the corrected temperature; The raw temperature collected by the infrared sensor; The ambient temperature; k This is an environmental correction factor; The formula for determining eddy current heating is as follows:

[0035] In the formula, Temperature rise in equipment components caused by eddy current effect; I This is the rated current of the line; d The outer diameter of the line; The formula for determining heat generation due to poor contact is as follows:

[0036] In the formula, The contact resistance of the line or plug; This represents the temperature rise difference between the plug and the wiring. This represents the thermal resistance of the corresponding component.

[0037] As a preferred solution for energy storage equipment detection and diagnosis device based on multi-model fusion and AI collaboration, in the single-dimensional model diagnosis unit, after the pressure detection module collects data, the pressure test model performs liquid cooling system leakage diagnosis through pressure change rate formula and leakage estimation formula; and performs liquid cooling system low liquid diagnosis through low liquid judgment formula and optimal liquid replenishment formula. The formula for the rate of change of pressure is:

[0038] In the formula, The pressure change rate of the liquid cooling system; , The measured pressures of the liquid cooling system at different times; , For the corresponding , The detection time; The formula for estimating the leakage amount is:

[0039] In the formula, This refers to the amount of leakage. V Volume of the liquid cooling system; The formula for determining low fluid volume is:

[0040] In the formula, h Real-time liquid level of the liquid cooling system; This represents the initial liquid level of the liquid cooling system. This represents the theoretical change in liquid level. A This represents the cross-sectional area of ​​the liquid storage chamber in the liquid cooling system. The formula for the optimal fluid replacement volume is:

[0041] In the formula, This represents the optimal replenishment volume for the liquid cooling system. This is the rated optimal liquid level for the liquid cooling system.

[0042] As a preferred solution for a detection and diagnosis device for energy storage equipment based on multi-model fusion and AI collaboration, in the single-dimensional model diagnosis unit, after the active balancing battery detection module collects data, the active balancing battery management system model calculates the single-dimensional diagnosis result through the battery consistency evaluation formula, the battery degradation diagnosis formula, and the optimal balancing current formula. The formula for evaluating battery consistency is as follows:

[0043] In the formula, For battery pack SOC balance; This represents the maximum SOC (State of Charge) of a single cell in the battery pack. This represents the minimum SOC (State of Charge) of a single cell in the battery pack. This represents the average SOC of each cell in the battery pack. The battery degradation diagnosis formula is as follows:

[0044] In the formula, For the first i The internal resistance decay rate of each battery cell; For the first i The current measured internal resistance of each individual battery cell; For the first i The initial internal resistance of each battery cell; The formula for the optimal equilibrium current is:

[0045] In the formula, For the first i The active balancing current of each battery cell; This is the equilibrium coefficient.

[0046] As a preferred solution for a detection and diagnosis device for energy storage equipment based on multi-model fusion and AI collaboration, in the multi-model fusion diagnosis unit, during the process of obtaining the comprehensive diagnosis value through weighted fusion using the self-built core fusion diagnosis model, the weight of each diagnosis model is determined by a confidence formula; based on the weight of each diagnosis model, the comprehensive diagnosis value is calculated using a weighted fusion formula. The confidence level formula is as follows:

[0047] In the formula, For the first i The weights of each diagnostic model; For the first i Diagnostic accuracy of a diagnostic model; i =1 corresponds to the infrared temperature recognition model; i =2 corresponds to the image recognition model; i =3 corresponds to the active equalization battery management system model; i =4 corresponds to the EMS energy management system model; The weighted fusion formula is as follows:

[0048] In the formula, D This is a comprehensive diagnostic value; For the first i A single-dimensional diagnostic value output by a diagnostic model.

[0049] As a preferred solution for energy storage equipment testing and diagnostic devices based on multi-model fusion and AI collaboration, in the cross-manufacturer interface adaptation unit, when the AI ​​processing module parses the interface protocol, the adaptation effect is determined by the interface protocol matching degree formula. The formula for the interface protocol matching degree is:

[0050] In the formula, M P represents the protocol matching degree; P is the interface protocol feature vector of the energy storage device to be adapted. The first one pre-stored in the database j Interface protocol feature vectors of each manufacturer; In the optimal operation and maintenance plan generation unit, during the process of calculating and screening the globally optimal operation and maintenance plan using the comprehensive loss value formula, the comprehensive loss value formula is as follows:

[0051] In the formula, L This is the overall loss value; The estimated processing time for the operation and maintenance plan; This is the estimated processing cost for the operation and maintenance plan.

[0052] The present invention has the following advantages: First, by integrating multi-dimensional data for diagnosis, the comprehensiveness of the assessment is improved. Data from multiple types of equipment is collected simultaneously and weighted and integrated through a core fusion model, breaking through the limitations of a single dimension, effectively avoiding missed detection of related faults, and providing a more comprehensive reflection of the overall status of the equipment.

[0053] Secondly, it enhances versatility and efficiency through automatic cross-vendor adaptation. By automatically parsing interface protocols and generating adaptation code using AI, no manual development is required, shortening the adaptation cycle, adapting to scenarios where multiple manufacturers' devices coexist, and lowering the threshold for cross-brand operation and maintenance.

[0054] Third, it can adapt to extreme environments and improve ease of operation. It supports full voice command control, eliminating the need for manual contact with the equipment, adapting to harsh environments such as high altitudes and cold weather, avoiding the impact of low temperatures on operation, and reducing on-site operational risks.

[0055] Fourth, drive operational and maintenance decisions through quantitative methods to scientifically balance multiple demands. By using formula-based quantitative analysis and loss value calculation, the optimal operational and maintenance solution is selected to replace experience-based decisions, taking into account risk, timeliness, and cost, and reducing downtime losses.

[0056] Fifth, the model is dynamically iterated to continuously optimize diagnostic capabilities. Based on post-fault operational data, model parameters and weights are updated to adapt to changes in equipment status, forming a closed-loop optimization system to improve the relevance and reliability of subsequent diagnoses. Attached Figure Description

[0057] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0058] The structures, proportions, sizes, etc. illustrated in this specification are only for the purpose of assisting those skilled in the art in understanding and reading the content disclosed herein, and are not intended to limit the conditions under which the present invention can be implemented. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportions, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should still fall within the scope of the technical content disclosed in the present invention.

[0059] Figure 1 This is a flowchart illustrating the energy storage device detection and diagnosis method based on multi-model fusion and AI collaboration provided in Embodiment 1 of the present invention. Figure 2 This is a schematic diagram of the architecture of the energy storage device detection and diagnosis device based on multi-model fusion and AI collaboration provided in Embodiment 2 of the present invention. Detailed Implementation

[0060] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Example 1

[0061] See Figure 1 Embodiment 1 of the present invention provides a method for detecting and diagnosing energy storage devices based on multi-model fusion and AI collaboration, comprising the following steps: S1. Start the portable host and drive each detection module to complete self-test; the AI ​​processing module loads the preset diagnostic model and manufacturer interface protocol pre-training data; the communication module establishes the device connection foundation by scanning the surrounding energy storage devices; S2. Based on the device connection, the camera image recognition module, infrared temperature detection module, pressure detection module and active equalization battery detection module are controlled by voice commands to simultaneously collect images of the water pipe connection, device surface temperature, liquid cooling system pressure and level and battery pack individual parameter data to obtain multi-dimensional raw data. S3. Based on the multi-dimensional raw data, each diagnostic model is quantitatively calculated using corresponding mathematical formulas, and outputs single-dimensional diagnostic results and model confidence respectively. S4. Based on the single-dimensional diagnostic results and the model confidence level, a weighted fusion is performed using a self-built core fusion model to obtain a comprehensive diagnostic value; based on the comprehensive diagnostic value, the fault level and type are determined, and a comprehensive diagnostic result is generated. S5. Based on the supplementary data requirements of the comprehensive diagnostic results, the AI ​​processing module parses the target energy storage device interface protocol through an artificial intelligence code editor and automatically generates adaptation code; based on the adaptation code, it establishes communication with the target energy storage device and EMS system to collect supplementary auxiliary data. S6. Based on the comprehensive diagnostic results and the auxiliary data, the globally optimal operation and maintenance solution is calculated and screened using the comprehensive loss value formula. S7. Output the global optimal operation and maintenance plan to the display alarm module; based on the global optimal operation and maintenance plan, the testing personnel execute auxiliary operations through voice command-driven devices, and upload fault data, diagnostic process and execution status to the cloud platform; S8. Based on the equipment operation data after fault handling, update the preset parameters and confidence weight coefficients of the diagnostic model to improve the accuracy of subsequent diagnosis.

[0062] In this embodiment, in step S1, the portable host is started to drive each detection module to complete self-test; the AI ​​processing module loads the preset diagnostic model and manufacturer interface protocol pre-training data; and the communication module establishes the device connection basis by scanning the surrounding energy storage devices.

[0063] Specifically, after the portable host is started, it first sends self-test commands to various detection modules such as the camera image recognition module and the infrared temperature detection module. Each module then completes the validity verification of the hardware functions and returns the results. At the same time, the AI ​​processing module retrieves the preset diagnostic models such as infrared temperature recognition and image recognition from the local storage unit, as well as the pre-trained interface protocol feature data of mainstream energy storage manufacturers, and loads them. The communication module then starts the wired / wireless scanning mode, identifies the communication identifiers of surrounding energy storage devices, and establishes an initial connection link, laying the foundation for subsequent data interaction.

[0064] In this embodiment, in step S2, based on the device connection, the camera image recognition module, infrared temperature detection module, pressure detection module, and active equalization battery detection module are controlled by voice commands to simultaneously collect images of the water pipe connection, device surface temperature, liquid cooling system pressure and level, and battery pack individual parameter data to obtain multi-dimensional raw data. Specifically, based on the device connection established by S1, after the testing personnel issue a voice command, the voice interaction module first performs noise reduction and semantic parsing on the command, and then sends a synchronous acquisition command to the corresponding testing module; the camera image recognition module captures high-definition images of the water pipe joint area of ​​the line; the infrared temperature detection module performs non-contact temperature acquisition of key components of the equipment (lines, liquid cooling plates, etc.); the pressure detection module acquires the pressure and liquid level data of the liquid cooling system in real time; and the active equalization battery detection module synchronously collects parameters such as voltage and internal resistance of each cell in the battery pack. All data are preprocessed and summarized into multi-dimensional raw data and transmitted to the AI ​​processing module.

[0065] In this embodiment, in step S3, based on the multi-dimensional raw data, each diagnostic model performs quantitative calculations using corresponding mathematical formulas, and outputs single-dimensional diagnostic results and model confidence levels respectively.

[0066] Specifically, the AI ​​processing module distributes multi-dimensional raw data to corresponding diagnostic models according to their types: the image recognition model extracts and matches feature points in the water pipe connection images; the infrared temperature recognition model performs environmental correction and heat determination calculations on the equipment temperature data; the pressure test model analyzes and calculates the rate of change of pressure and liquid level data in the liquid cooling system; and the active equalization battery management model calculates the attenuation and equalization of battery cell parameters. After each model completes the quantitative calculation, it outputs a single-dimensional diagnostic result reflecting the corresponding dimension of the fault state, and outputs the model confidence level of the result based on the historical diagnostic accuracy.

[0067] Wherein, after the camera image recognition module collects data, the image recognition model calculates the single-dimensional diagnostic result by using a feature point matching similarity formula; The formula for feature point matching similarity is:

[0068] In the formula, S Match feature points with similarity; N This represents the number of feature points that successfully match the image to be detected with the standard image. This represents the total number of standard feature points corresponding to the connector.

[0069] When S < 75%, the connection of the line plug or water pipe joint is determined to be abnormal. Simultaneously, the location of the abnormality is determined using the deviation coordinates of characteristic points.

[0070] In this embodiment, after the infrared temperature detection module collects data, the infrared temperature recognition model calculates the single-dimensional diagnostic result by sequentially using the temperature correction formula, the eddy current heating determination formula, and the poor contact heating determination formula. The temperature correction formula is as follows:

[0071] In the formula, T This is the corrected temperature; The raw temperature collected by the infrared sensor; The ambient temperature; k This is an environmental correction factor; The formula for determining eddy current heating is as follows:

[0072] In the formula, Temperature rise in equipment components caused by eddy current effect; I This is the rated current of the line; d The outer diameter of the line; when When the temperature is greater than 5℃, eddy current heating is determined to be present.

[0073] The formula for determining heat generation due to poor contact is as follows:

[0074] In the formula, The contact resistance of the line or plug; This represents the temperature rise difference between the plug and the wiring. This represents the thermal resistance of the corresponding component. When... When the resistance is greater than 50mΩ, it is determined that there is poor contact and overheating.

[0075] In this embodiment, after the pressure detection module collects data, the pressure test model performs liquid cooling system leakage diagnosis using the pressure change rate formula and leakage estimation formula; and performs liquid cooling system low liquid diagnosis using the low liquid determination formula and optimal liquid replenishment formula. The formula for the rate of change of pressure is:

[0076] In the formula, The pressure change rate of the liquid cooling system; , The measured pressures of the liquid cooling system at different times; , For the corresponding , The detection time; when the absolute value of the pressure change rate is less than 0.05 MPa / min, it is determined that there is no leakage.

[0077] The formula for estimating the leakage amount is:

[0078] In the formula, This refers to the amount of leakage. V Volume of the liquid cooling system; The formula for determining low fluid volume is:

[0079] In the formula, h Real-time liquid level of the liquid cooling system; This represents the initial liquid level of the liquid cooling system. This represents the theoretical change in liquid level. A Let be the cross-sectional area of ​​the liquid storage chamber in the liquid cooling system; when h > At that time, it was determined that there was no insufficient fluid.

[0080] The formula for the optimal fluid replacement volume is:

[0081]

[0082] In the formula, This represents the optimal replenishment volume for the liquid cooling system. This is the rated optimal liquid level for the liquid cooling system; This represents the minimum liquid level in the liquid cooling system. This represents the maximum liquid level in the liquid cooling system.

[0083] In this embodiment, after the active balancing battery detection module collects data, the active balancing battery management system model calculates the single-dimensional diagnostic result using the battery consistency evaluation formula, the battery degradation diagnosis formula, and the optimal balancing current formula. The formula for evaluating battery consistency is as follows:

[0084] In the formula, For battery pack SOC balance; This represents the maximum SOC (State of Charge) of a single cell in the battery pack. This represents the minimum SOC (State of Charge) of a single cell in the battery pack. This represents the average SOC of each cell in the battery pack; when When the value is greater than 0.9, the consistency is considered normal.

[0085] The battery degradation diagnosis formula is as follows:

[0086] In the formula, For the first i The internal resistance decay rate of each battery cell; For the first i The current measured internal resistance of each individual battery cell; For the first i The initial internal resistance of each battery cell; when A reading >0.2 indicates severe battery degradation.

[0087] The formula for the optimal equilibrium current is:

[0088] In the formula, For the first i The active balancing current of each battery cell; This is the equilibrium coefficient.

[0089] In this embodiment, in step S4, based on the single-dimensional diagnostic results and the model confidence, a weighted fusion is performed through a self-built core fusion model to obtain a comprehensive diagnostic value; based on the comprehensive diagnostic value, the fault level and type are determined, and a comprehensive diagnostic result is generated.

[0090] Specifically, the AI ​​processing module first calculates the weight ratio of each model based on the confidence level of each model using a weighting formula; then it performs a weighted summation of the single-dimensional diagnostic results with their corresponding weights to obtain a comprehensive diagnostic value; subsequently, it calls the preset fault level threshold rules and determines the current fault level of the equipment based on the numerical range of the comprehensive diagnostic value, such as below 1 being normal and 1-3 being slightly abnormal. At the same time, it integrates the abnormality types of each single-dimensional diagnostic result to generate a comprehensive diagnostic result that includes the fault location and related components.

[0091] In the process of obtaining the comprehensive diagnostic value through weighted fusion using the self-built core fusion diagnostic model, the weight of each diagnostic model is determined by a confidence formula; based on the weight of each diagnostic model, the comprehensive diagnostic value is calculated using a weighted fusion formula. The confidence level formula is as follows:

[0092] In the formula, For the first i The weights of each diagnostic model; For the first i Diagnostic accuracy of a diagnostic model; i =1 corresponds to the infrared temperature recognition model; i =2 corresponds to the image recognition model; i =3 corresponds to the active equalization battery management system model; i =4 corresponds to the EMS energy management system model; The weighted fusion formula is as follows:

[0093] In the formula, D This is a comprehensive diagnostic value; For the first i A single-dimensional diagnostic value output by a diagnostic model.

[0094] In this embodiment, in step S5, based on the supplementary data requirements of the comprehensive diagnostic results, the AI ​​processing module parses the target energy storage device interface protocol through an artificial intelligence code editor and automatically generates adaptation code; based on the adaptation code, it establishes communication with the target energy storage device and the EMS system to collect supplementary auxiliary data.

[0095] Specifically, when the comprehensive diagnostic results require supplementary information such as EMS system data, the AI ​​processing module first extracts the interface protocol features of the target energy storage device and performs matching calculations with the pre-trained manufacturer protocol library through an artificial intelligence code editor. After the matching is completed, the adaptation code is automatically generated to establish a communication link between the detection device and the target energy storage device and EMS system. Subsequently, a data request command is sent to the EMS system to collect auxiliary data such as equipment operating power and historical fault records, and these are added to the dataset corresponding to the comprehensive diagnostic results.

[0096] Specifically, when the AI ​​processing module parses the interface protocol, the adaptation effect is determined by the interface protocol matching degree formula. The formula for the interface protocol matching degree is:

[0097] In the formula, M P represents the protocol matching degree; P is the interface protocol feature vector of the energy storage device to be adapted. The first one pre-stored in the database j Interface protocol feature vectors of each manufacturer; In this embodiment, in step S6, based on the comprehensive diagnostic results and the auxiliary data, the globally optimal operation and maintenance solution is calculated and screened using the comprehensive loss value formula.

[0098] Specifically, the AI ​​processing module integrates the diagnostic results with the supplementary auxiliary data and substitutes them into the comprehensive loss value formula: the comprehensive diagnostic value reflects the fault risk, the estimated processing time reflects the efficiency cost, and the estimated resource input reflects the economic cost. The comprehensive loss value is calculated for multiple candidate operation and maintenance solutions. Finally, the solution with the smallest comprehensive loss value is selected as the globally optimal operation and maintenance solution that takes into account risk, time, and cost.

[0099] In the process of calculating and selecting the globally optimal operation and maintenance solution using the comprehensive loss value formula, the comprehensive loss value formula is as follows:

[0100] In the formula, L This is the overall loss value; The estimated processing time for the operation and maintenance plan; This is the estimated processing cost for the operation and maintenance plan.

[0101] In this embodiment, in step S7, the globally optimal operation and maintenance plan is output to the display alarm module; based on the globally optimal operation and maintenance plan, the testing personnel execute auxiliary operations through voice command-driven devices, and upload fault data, diagnostic process and execution status to the cloud platform.

[0102] Specifically, the AI ​​processing module sends the globally optimal operation and maintenance plan, including fault description, operation steps, and precautions, to the display and alarm module for visualization on a high-definition touchscreen. After the inspection personnel confirm the plan, they drive the device to perform auxiliary operations through voice commands, such as locating faulty components and outputting operation instructions. At the same time, the communication module uploads fault data, diagnostic process logs for each step, and operation execution status to the cloud platform in real time, realizing remote data storage and synchronization.

[0103] In this embodiment, in step S8, based on the equipment operation data after fault handling, the preset parameters and confidence weight coefficients of the diagnostic model are updated to improve the accuracy of subsequent diagnosis.

[0104] Specifically, after the fault handling is completed, the detection device continues to collect subsequent operating data of the equipment. The AI ​​processing module compares this data with the diagnostic results and operation and maintenance plan before processing, and analyzes the prediction deviation of each diagnostic model. Then, based on the degree of deviation, the preset parameters of the diagnostic model are adjusted, and the confidence weight coefficient of each model is updated to optimize the weight ratio of each model in the subsequent diagnosis process and improve the fit and accuracy of the diagnostic results.

[0105] The application scenarios of this invention are as follows: In the operation and maintenance of outdoor energy storage power stations in cold northern regions, this invention enables contactless operation through voice interaction, avoiding the impact of low temperatures on manual operation and efficiently completing multi-dimensional equipment testing and fault diagnosis.

[0106] In the scenario of distributed energy storage cabinet cluster management where multiple manufacturers' energy storage devices coexist, this invention automatically adapts to the interface protocols of different manufacturers, eliminating the need to repeatedly develop adaptation code and quickly achieving unified detection of devices across manufacturers.

[0107] In the daily inspection of large-scale energy storage power stations with liquid cooling systems as the main component, this invention simultaneously collects pressure, liquid level and temperature data, analyzes leakage and low liquid problems through quantitative models, and outputs appropriate liquid replenishment solutions and operation and maintenance guidelines.

[0108] In the scenario of health status assessment after long-term operation of energy storage battery packs, this invention analyzes the voltage, internal resistance and SOC data of individual battery cells to determine the battery degradation and balance status, and provides targeted balance strategies and maintenance suggestions.

[0109] In emergency response scenarios involving sudden failures of energy storage equipment, this invention rapidly integrates multi-dimensional detection data to generate comprehensive diagnostic results, selects the optimal operation and maintenance solution, shortens fault response and handling time, and reduces downtime losses.

[0110] In the routine testing scenarios of energy storage systems used for power grid peak shaving, this invention integrates equipment operation data and EMS system information to output an operation and maintenance solution that balances efficiency and cost, ensuring that the energy storage system can stably respond to the power grid dispatching requirements.

[0111] It should be noted that the method of this disclosure embodiment can be executed by a single device, such as a computer or server. The method of this embodiment can also be applied to a distributed scenario, where multiple devices cooperate to complete the task. In such a distributed scenario, one of these devices may execute only one or more steps of the method of this disclosure embodiment, and the multiple devices will interact with each other to complete the method described.

[0112] It should be noted that the above description describes some embodiments of this disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the above embodiments and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous. Example 2

[0113] See Figure 2 Embodiment 2 of the present invention also provides an energy storage device detection and diagnosis device based on multi-model fusion and AI collaboration, comprising: The power-on initialization unit 001 is used to start the portable host and drive each detection module to complete self-test; the AI ​​processing module loads the preset diagnostic model and pre-trained data of the manufacturer's interface protocol; the communication module establishes the device connection foundation by scanning the surrounding energy storage devices. The multi-dimensional data acquisition unit 002 is used to control the camera image recognition module, infrared temperature detection module, pressure detection module and active equalization battery detection module through voice commands based on the device connection foundation, and simultaneously acquire the line water pipe connection image, equipment surface temperature, liquid cooling system pressure and liquid level and battery pack individual parameter data to obtain multi-dimensional raw data. The single-dimensional model diagnostic unit 003 is used to perform quantitative calculations on each of the diagnostic models based on the multi-dimensional raw data and corresponding mathematical formulas, and output single-dimensional diagnostic results and model confidence respectively. The multi-model fusion diagnostic unit 004 is used to perform weighted fusion based on the single-dimensional diagnostic results and the model confidence level through a self-built core fusion model to obtain a comprehensive diagnostic value; based on the comprehensive diagnostic value, the fault level and type are determined, and a comprehensive diagnostic result is generated. The cross-manufacturer interface adaptation unit 005 is used to supplement data requirements based on the comprehensive diagnostic results. The AI ​​processing module parses the target energy storage device interface protocol through an artificial intelligence code editor and automatically generates adaptation code. Based on the adaptation code, communication is established with the target energy storage device and EMS system to supplement and collect auxiliary data. The optimal operation and maintenance solution generation unit 006 is used to calculate and screen the globally optimal operation and maintenance solution based on the comprehensive diagnostic results and the auxiliary data, using the comprehensive loss value formula. The solution execution and data upload unit 007 is used to output the global optimal operation and maintenance solution to the display alarm module; based on the global optimal operation and maintenance solution, the testing personnel execute auxiliary operations through voice command-driven devices and upload fault data, diagnostic process and execution status to the cloud platform. The model optimization and update unit 008 is used to update the preset parameters and confidence weight coefficients of the diagnostic model based on the equipment operation data after fault handling, so as to improve the accuracy of subsequent diagnosis.

[0114] The hardware configuration of this invention is as follows: The portable host unit features a waterproof and dustproof lightweight shell with an IP65 protection rating; the camera image recognition module uses a 12-megapixel industrial camera, supporting low-light and backlight environment acquisition; the infrared temperature detection module has a temperature measurement range of -30℃ to 200℃ and an accuracy of ±0.3℃; the pressure detection module has a range of 0~2MPa and an accuracy of ±0.01MPa; the active balancing battery detection module supports simultaneous acquisition of single-cell voltage and internal resistance; the AI ​​processing module uses a Qualcomm Snapdragon 8 Gen3 processor, with a built-in GPT-4 Turbo large language model, pre-trained with interface protocols from more than 20 mainstream manufacturers such as CATL, BYD, and Sungrow Power; the voice interaction module has a recognition accuracy of ≥96% and supports operation at -30℃; the communication module supports 5G full-band and Bluetooth 5.3.

[0115] In this embodiment, the diagnostic model in the power-on initialization unit 001 includes: an infrared temperature recognition model, an image recognition model, a stress test model, an active equalization battery management system model, and an EMS energy management system model.

[0116] In this embodiment, in the single-dimensional model diagnostic unit 003, after the camera image recognition module collects data, the image recognition model calculates the single-dimensional diagnostic result through the feature point matching similarity formula. The formula for feature point matching similarity is:

[0117] In the formula, S Match feature points with similarity; N This represents the number of feature points that successfully match the image to be detected with the standard image. This represents the total number of standard feature points corresponding to the connector.

[0118] In this embodiment, in the single-dimensional model diagnostic unit 003, after the infrared temperature detection module collects data, the infrared temperature recognition model calculates the single-dimensional diagnostic result by sequentially using the temperature correction formula, the eddy current heating determination formula, and the poor contact heating determination formula. The temperature correction formula is as follows:

[0119] In the formula, T This is the corrected temperature; The raw temperature collected by the infrared sensor; The ambient temperature; k This is an environmental correction factor; The formula for determining eddy current heating is as follows:

[0120] In the formula, Temperature rise in equipment components caused by eddy current effect; I This is the rated current of the line; d The outer diameter of the line; The formula for determining heat generation due to poor contact is as follows:

[0121] In the formula, The contact resistance of the line or plug; This represents the temperature rise difference between the plug and the wiring. This represents the thermal resistance of the corresponding component.

[0122] In this embodiment, in the single-dimensional model diagnostic unit 003, after the pressure detection module collects data, the pressure test model performs liquid cooling system leakage diagnosis through the pressure change rate formula and leakage estimation formula; and performs liquid cooling system low liquid diagnosis through the low liquid determination formula and optimal liquid replenishment formula. The formula for the rate of change of pressure is:

[0123] In the formula, The pressure change rate of the liquid cooling system; , The measured pressures of the liquid cooling system at different times; , For the corresponding , The detection time; The formula for estimating the leakage amount is:

[0124] In the formula, This refers to the amount of leakage. V Volume of the liquid cooling system; The formula for determining low fluid volume is:

[0125] In the formula, h Real-time liquid level of the liquid cooling system; This represents the initial liquid level of the liquid cooling system. This represents the theoretical change in liquid level. A This represents the cross-sectional area of ​​the liquid storage chamber in the liquid cooling system. The formula for the optimal fluid replacement volume is:

[0126] In the formula, This represents the optimal replenishment volume for the liquid cooling system. This is the rated optimal liquid level for the liquid cooling system.

[0127] In this embodiment, in the single-dimensional model diagnostic unit 003, after the active balancing battery detection module collects data, the active balancing battery management system model calculates the single-dimensional diagnostic result through the battery consistency evaluation formula, the battery degradation diagnostic formula, and the optimal balancing current formula. The formula for evaluating battery consistency is as follows:

[0128] In the formula, For battery pack SOC balance; This represents the maximum SOC (State of Charge) of a single cell in the battery pack. This represents the minimum SOC (State of Charge) of a single cell in the battery pack. This represents the average SOC of each cell in the battery pack. The battery degradation diagnosis formula is as follows:

[0129] In the formula, For the first i The internal resistance decay rate of each battery cell; For the first i The current measured internal resistance of each individual battery cell; For the first iThe initial internal resistance of each battery cell; The formula for the optimal equilibrium current is:

[0130] In the formula, For the first i The active balancing current of each battery cell; This is the equilibrium coefficient.

[0131] In this embodiment, in the multi-model fusion diagnostic unit 004, during the process of obtaining the comprehensive diagnostic value through weighted fusion using the self-built core fusion diagnostic model, the weight of each diagnostic model is determined by a confidence formula; based on the weight of each diagnostic model, the comprehensive diagnostic value is calculated using a weighted fusion formula. The confidence level formula is as follows:

[0132] In the formula, For the first i The weights of each diagnostic model; For the first i Diagnostic accuracy of a diagnostic model; i =1 corresponds to the infrared temperature recognition model; i =2 corresponds to the image recognition model; i =3 corresponds to the active equalization battery management system model; i =4 corresponds to the EMS energy management system model; The weighted fusion formula is as follows:

[0133] In the formula, D This is a comprehensive diagnostic value; For the first i A single-dimensional diagnostic value output by a diagnostic model.

[0134] In this embodiment, in the cross-manufacturer interface adaptation unit 005, when the AI ​​processing module parses the interface protocol, the adaptation effect is determined by the interface protocol matching degree formula. The formula for the interface protocol matching degree is:

[0135] In the formula, M P represents the protocol matching degree; P is the interface protocol feature vector of the energy storage device to be adapted. The first one pre-stored in the database j Interface protocol feature vectors of each manufacturer; In the optimal operation and maintenance solution generation unit 006, during the process of calculating and screening the globally optimal operation and maintenance solution using the comprehensive loss value formula, the comprehensive loss value formula is as follows:

[0136] In the formula, L This is the overall loss value; The estimated processing time for the operation and maintenance plan; This is the estimated processing cost for the operation and maintenance plan.

[0137] It should be noted that the information interaction and execution process between the modules of the above system are based on the same concept as the method embodiment in Embodiment 1 of this application, and the resulting technical effects are the same as those in the method embodiment of this application. For details, please refer to the description in the method embodiment shown above in this application, and it will not be repeated here. Example 3

[0138] Embodiment 3 of the present invention provides a non-transitory computer-readable storage medium storing program code of a method for detecting and diagnosing energy storage devices based on multi-model fusion and AI collaboration. The program code includes instructions for executing the method for detecting and diagnosing energy storage devices based on multi-model fusion and AI collaboration as described in Embodiment 1 or any possible implementation thereof.

[0139] Computer-readable storage media can be any available medium that a computer can access, or a data storage device such as a server or data center that integrates one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)). Example 4

[0140] Embodiment 4 of the present invention provides an electronic device, including: a memory and a processor; The processor and the memory communicate with each other via a bus; the memory stores program instructions that can be executed by the processor, and the processor can call the program instructions to execute the energy storage device detection and diagnosis method based on multi-model fusion and AI collaboration in Embodiment 1 or any possible implementation thereof.

[0141] Specifically, a processor can be implemented in hardware or software. When implemented in hardware, the processor can be a logic circuit, an integrated circuit, etc. When implemented in software, the processor can be a general-purpose processor that reads software code stored in memory. This memory can be integrated into the processor or located outside the processor and exist independently.

[0142] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable system. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means.

[0143] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing systems. They can be centralized on a single computing system or distributed across a network of multiple computing systems. Optionally, they can be implemented using program code executable by a computing system, thereby storing them in a storage system for execution by the computing system. In some cases, the steps shown or described can be performed in a different order than those presented herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0144] Although the present invention has been described in detail above with general descriptions and specific embodiments, modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, all such modifications or improvements made without departing from the spirit of the present invention fall within the scope of protection claimed by the present invention.

Claims

1. A detection and diagnosis method for energy storage devices based on multi-model fusion and AI collaboration, characterized in that, include: The portable host is started, driving each detection module to complete self-test; the AI ​​processing module loads the preset diagnostic model and pre-trained data of the manufacturer's interface protocol; The communication module establishes a basic connection between the devices by scanning nearby energy storage devices; Based on the aforementioned device connectivity, the camera image recognition module, infrared temperature detection module, pressure detection module, and active equalization battery detection module are controlled via voice commands to simultaneously collect images of the water pipe connection, device surface temperature, liquid cooling system pressure and level, and individual battery parameters, thereby obtaining multi-dimensional raw data. Based on the multi-dimensional raw data, each diagnostic model is quantitatively calculated using corresponding mathematical formulas, and outputs single-dimensional diagnostic results and model confidence respectively. Based on the single-dimensional diagnostic results and the model confidence, a comprehensive diagnostic value is obtained by weighted fusion through a self-built core fusion model. Based on the comprehensive diagnostic values, the fault level and type are determined, and a comprehensive diagnostic result is generated. Based on the supplementary data requirements of the comprehensive diagnostic results, the AI ​​processing module parses the target energy storage device interface protocol through an artificial intelligence code editor and automatically generates adaptation code; based on the adaptation code, it establishes communication with the target energy storage device and EMS system to collect supplementary auxiliary data. Based on the comprehensive diagnostic results and the auxiliary data, the globally optimal operation and maintenance solution is calculated and screened using the comprehensive loss value formula. The globally optimal operation and maintenance plan is output to the display and alarm module; based on the globally optimal operation and maintenance plan, the testing personnel execute auxiliary operations through voice command-driven devices, and upload fault data, diagnostic process and execution status to the cloud platform; Based on the equipment operation data after fault handling, the preset parameters and confidence weight coefficients of the diagnostic model are updated to improve the accuracy of subsequent diagnosis.

2. The energy storage device detection and diagnosis method based on multi-model fusion and AI collaboration according to claim 1, characterized in that, The diagnostic models include: infrared temperature recognition model, image recognition model, stress test model, active equalization battery management system model, and EMS energy management system model.

3. The energy storage device detection and diagnosis method based on multi-model fusion and AI collaboration according to claim 2, characterized in that, After the camera image recognition module collects data, the image recognition model calculates the single-dimensional diagnostic result using a feature point matching similarity formula. The formula for feature point matching similarity is: ； In the formula, S Match feature points with similarity; N This represents the number of feature points that successfully match the image to be detected with the standard image. This represents the total number of standard feature points corresponding to the connector.

4. The energy storage device detection and diagnosis method based on multi-model fusion and AI collaboration according to claim 3, characterized in that, After the infrared temperature detection module collects data, the infrared temperature recognition model calculates the single-dimensional diagnostic result by sequentially using the temperature correction formula, the eddy current heating determination formula, and the poor contact heating determination formula. The temperature correction formula is as follows: ； In the formula, T This is the corrected temperature; The raw temperature collected by the infrared sensor; The ambient temperature; k This is an environmental correction factor; The formula for determining eddy current heating is as follows: ； In the formula, Temperature rise in equipment components caused by eddy current effect; I This is the rated current of the line; d The outer diameter of the line; The formula for determining heat generation due to poor contact is as follows: ； In the formula, The contact resistance of the line or plug; This represents the temperature rise difference between the plug and the wiring. This represents the thermal resistance of the corresponding component.

5. The energy storage device detection and diagnosis method based on multi-model fusion and AI collaboration according to claim 4, characterized in that, After the pressure detection module collects data, the pressure test model diagnoses liquid leakage in the liquid cooling system using the pressure change rate formula and the leakage estimation formula; it also diagnoses low liquid levels in the liquid cooling system using the low liquid determination formula and the optimal liquid replenishment formula. The formula for the rate of change of pressure is: ； In the formula, The pressure change rate of the liquid cooling system; , The measured pressures of the liquid cooling system at different times; , For the corresponding , The detection time; The formula for estimating the leakage amount is: ； In the formula, This refers to the amount of leakage. V Volume of the liquid cooling system; The formula for determining low fluid volume is: ； In the formula, h Real-time liquid level of the liquid cooling system; This represents the initial liquid level of the liquid cooling system. This represents the theoretical change in liquid level. A This represents the cross-sectional area of ​​the liquid storage chamber in the liquid cooling system. The formula for the optimal fluid replacement volume is: ； In the formula, This represents the optimal replenishment volume for the liquid cooling system. This is the rated optimal liquid level for the liquid cooling system.

6. The energy storage device detection and diagnosis method based on multi-model fusion and AI collaboration according to claim 5, characterized in that, After the active balancing battery detection module collects data, the active balancing battery management system model calculates the single-dimensional diagnostic result using the battery consistency evaluation formula, the battery degradation diagnosis formula, and the optimal balancing current formula. The formula for evaluating battery consistency is as follows: ； In the formula, For battery pack SOC balance; This represents the maximum SOC (State of Charge) of a single cell in the battery pack. This represents the minimum SOC (State of Charge) of a single cell in the battery pack. This represents the average SOC of each cell in the battery pack. The battery degradation diagnosis formula is as follows: ； In the formula, For the first i The internal resistance decay rate of each battery cell; For the first i The current measured internal resistance of each individual battery cell; For the first i The initial internal resistance of each battery cell; The formula for the optimal equilibrium current is: ； In the formula, For the first i The active balancing current of each battery cell; This is the equilibrium coefficient.

7. The energy storage device detection and diagnosis method based on multi-model fusion and AI collaboration according to claim 6, characterized in that, In the process of obtaining the comprehensive diagnostic value through weighted fusion using the self-built core fusion diagnostic model, the weight of each diagnostic model is determined by a confidence formula; based on the weight of each diagnostic model, the comprehensive diagnostic value is calculated using a weighted fusion formula. The confidence level formula is as follows: ； In the formula, For the first i The weights of each diagnostic model; For the first i Diagnostic accuracy of a diagnostic model; i =1 corresponds to the infrared temperature recognition model; i =2 corresponds to the image recognition model; i =3 corresponds to the active equalization battery management system model; i =4 corresponds to the EMS energy management system model; The weighted fusion formula is as follows: ； In the formula, D This is a comprehensive diagnostic value; For the first i A single-dimensional diagnostic value output by a diagnostic model.

8. The energy storage device detection and diagnosis method based on multi-model fusion and AI collaboration according to claim 7, characterized in that, When the AI ​​processing module parses the interface protocol, it determines the adaptation effect through the interface protocol matching degree formula. The formula for the interface protocol matching degree is: ； In the formula, M P represents the protocol matching degree; P is the interface protocol feature vector of the energy storage device to be adapted. The first one pre-stored in the database j Interface protocol feature vectors of each manufacturer; In the process of calculating and selecting the globally optimal operation and maintenance solution using the comprehensive loss value formula, the comprehensive loss value formula is as follows: ； In the formula, L This is the overall loss value; The estimated processing time for the operation and maintenance plan; This is the estimated processing cost for the operation and maintenance plan.

9. A detection and diagnosis device for energy storage equipment based on multi-model fusion and AI collaboration, employing the detection and diagnosis method for energy storage equipment based on multi-model fusion and AI collaboration as described in any one of claims 1-8, characterized in that, include: The power-on initialization unit is used to start the portable host and drive each detection module to complete self-test; the AI ​​processing module loads the preset diagnostic model and pre-trained data of the manufacturer's interface protocol; the communication module establishes the device connection foundation by scanning the surrounding energy storage devices. The multi-dimensional data acquisition unit is used to control the camera image recognition module, infrared temperature detection module, pressure detection module and active equalization battery detection module through voice commands based on the device connection foundation, and simultaneously acquire the line water pipe connection image, equipment surface temperature, liquid cooling system pressure and liquid level and battery pack individual parameter data to obtain multi-dimensional raw data; The single-dimensional model diagnostic unit is used to perform quantitative calculations on each of the diagnostic models based on the multi-dimensional raw data using corresponding mathematical formulas, and output single-dimensional diagnostic results and model confidence scores respectively. The multi-model fusion diagnostic unit is used to perform weighted fusion based on the single-dimensional diagnostic results and the model confidence level, and obtain a comprehensive diagnostic value by building a core fusion model. Based on the comprehensive diagnostic values, the fault level and type are determined, and a comprehensive diagnostic result is generated. A cross-manufacturer interface adaptation unit is used to supplement data requirements based on the comprehensive diagnostic results. The AI ​​processing module parses the target energy storage device interface protocol through an artificial intelligence code editor and automatically generates adaptation code. Based on the adaptation code, communication is established with the target energy storage device and EMS system to supplement and collect auxiliary data. The optimal operation and maintenance plan generation unit is used to calculate and screen the globally optimal operation and maintenance plan based on the comprehensive diagnostic results and the auxiliary data, using the comprehensive loss value formula. The solution execution and data upload unit is used to output the global optimal operation and maintenance solution to the display and alarm module; based on the global optimal operation and maintenance solution, the testing personnel execute auxiliary operations through voice command-driven devices and upload fault data, diagnostic process and execution status to the cloud platform. The model optimization and update unit is used to update the preset parameters and confidence weight coefficients of the diagnostic model based on the equipment operation data after fault handling, so as to improve the accuracy of subsequent diagnosis.

10. The energy storage device detection and diagnosis device based on multi-model fusion and AI collaboration according to claim 9, characterized in that, In the power-on initialization unit, the diagnostic model includes: an infrared temperature recognition model, an image recognition model, a stress test model, an active equalization battery management system model, and an EMS energy management system model. In the single-dimensional model diagnostic unit, after the camera image recognition module collects data, the image recognition model calculates the single-dimensional diagnostic result through a feature point matching similarity formula. The formula for feature point matching similarity is: ； In the formula, S Match feature points with similarity; N This represents the number of feature points that successfully match the image to be detected with the standard image. This represents the total number of standard feature points corresponding to the joint; In the single-dimensional model diagnostic unit, after the infrared temperature detection module collects data, the infrared temperature recognition model calculates the single-dimensional diagnostic result by sequentially using the temperature correction formula, the eddy current heating determination formula, and the poor contact heating determination formula. The temperature correction formula is as follows: ； In the formula, T This is the corrected temperature; The raw temperature collected by the infrared sensor; The ambient temperature; k This is an environmental correction factor; The formula for determining eddy current heating is as follows: ； In the formula, Temperature rise in equipment components caused by eddy current effect; I This is the rated current of the line; d The outer diameter of the line; The formula for determining heat generation due to poor contact is as follows: ； In the formula, The contact resistance of the line or plug; This represents the temperature rise difference between the plug and the wiring. The thermal resistance of the corresponding component; In the single-dimensional model diagnostic unit, after the pressure detection module collects data, the pressure test model performs liquid cooling system leakage diagnosis through the pressure change rate formula and leakage estimation formula; and performs liquid cooling system low liquid diagnosis through the low liquid determination formula and optimal liquid replenishment formula. The formula for the rate of change of pressure is: ； In the formula, The pressure change rate of the liquid cooling system; , The measured pressures of the liquid cooling system at different times; , For the corresponding , The detection time; The formula for estimating the leakage amount is: ； In the formula, This refers to the amount of leakage. V Volume of the liquid cooling system; The formula for determining low fluid volume is: ； In the formula, h Real-time liquid level of the liquid cooling system; This represents the initial liquid level of the liquid cooling system. This represents the theoretical change in liquid level. A This represents the cross-sectional area of ​​the liquid storage chamber in the liquid cooling system. The formula for the optimal fluid replacement volume is: ； In the formula, This represents the optimal replenishment volume for the liquid cooling system. This is the rated optimal liquid level for the liquid cooling system; In the single-dimensional model diagnostic unit, after the active balancing battery detection module collects data, the active balancing battery management system model calculates the single-dimensional diagnostic result through the battery consistency evaluation formula, the battery degradation diagnostic formula, and the optimal balancing current formula. The formula for evaluating battery consistency is as follows: ； In the formula, For battery pack SOC balance; This represents the maximum SOC (State of Charge) of a single cell in the battery pack. This represents the minimum SOC (State of Charge) of a single cell in the battery pack. This represents the average SOC of each cell in the battery pack. The battery degradation diagnosis formula is as follows: ； In the formula, For the first i The internal resistance decay rate of each battery cell; For the first i The current measured internal resistance of each individual battery cell; For the first i The initial internal resistance of each battery cell; The formula for the optimal equilibrium current is: ； In the formula, For the first i The active balancing current of each battery cell; This is the equilibrium coefficient; In the multi-model fusion diagnostic unit, during the process of obtaining the comprehensive diagnostic value through weighted fusion using the self-built core fusion diagnostic model, the weight of each diagnostic model is determined by a confidence formula; based on the weight of each diagnostic model, the comprehensive diagnostic value is calculated using a weighted fusion formula. The confidence level formula is as follows: ； In the formula, For the first i The weights of each diagnostic model; For the first i Diagnostic accuracy of a diagnostic model; i =1 corresponds to the infrared temperature recognition model; i =2 corresponds to the image recognition model; i =3 corresponds to the active equalization battery management system model; i =4 corresponds to the EMS energy management system model; The weighted fusion formula is as follows: ； In the formula, D This is a comprehensive diagnostic value; For the first i A single-dimensional diagnostic value output by a diagnostic model; In the cross-vendor interface adaptation unit, when the AI ​​processing module parses the interface protocol, the adaptation effect is determined by the interface protocol matching degree formula. The formula for the interface protocol matching degree is: ； In the formula, M P represents the protocol matching degree; P is the interface protocol feature vector of the energy storage device to be adapted. The first one pre-stored in the database j Interface protocol feature vectors of each manufacturer; In the optimal operation and maintenance plan generation unit, during the process of calculating and screening the globally optimal operation and maintenance plan using the comprehensive loss value formula, the comprehensive loss value formula is as follows: ； In the formula, L This is the overall loss value; The estimated processing time for the operation and maintenance plan; This is the estimated processing cost for the operation and maintenance plan.

Images